/*
CODE SHEET FOR FRAMINGHAM DATA FILE: 5209 RECORDS

Var Name Description Range/Codes
-------------------------------------------------------------
 1 lexam Last exam      2-16 for 1997 subjects who died
                             just prior to last exam

                        16   for 3218 subjects who were alive 
                             at exam 16

 2 surv Survival         0 = alive at exam 16
                         1 = died prior to exam 16

 3 cause Cause of death  0 = still alive
                         1 = sudden death from CHD
                         2 = other coronary heart disease
                         3 = stroke (CVA)
                         4 = other cerebral vascular disease
                         5 = cancer
                         6 = other causes of death
                         9 = cause unknown

 4 cexam Exam 1st CHD  1-16: exam at which CHD 1st diagnosed
                         0 = CHD never diagnosed

 5 chd Death from CHD    1 = Death from cause 1 or 2
                         0 = Alive or death from other cause

 6 cva Death from CVD    1 = Death from cause 3 or 4
                         0 = Alive or death from other cause

 7 ca Death from CA      1 = Death from cause 5
                         0 = Alive or death from other cause

 8 oth Death from OTH    1 = Death from cause 3-9
                         0 = Alive or death from CHD

 9 sex Sex               1 = Male
                         2 = Female

10 age Age              28-62 age (years) at first exam

11 ht Height            51.5 - 76.5 height (inches) 1st exam
                             6 missing values (-1)

12 wt Weight            67 - 300 weight (pounds) 1st exam
                             6 missing values (-1)

13 scl1 SC Exam 1       96 - 503 serum cholesterol (mg/100ml)
                             2,037 missing values (-1)*

14 scl2 SC Exam 2      115 - 568 serum cholesterol (mg/100ml)
                               626 missing values (-1)*

15 dbp Diastolic        50 - 160 mm/Hg 1st exam 
                              Blood Pressure

16 sbp Systolic         82 - 300 mm/Hg 1st exam
                              Blood Pressure

17 mrw Metropolitan     67 - 238 percent of "normal"
                              Relative Weight
                               6 missing values (-1)

18 smok Amount smoked    0 - 60 cigarettes per day (exam 1)
                               36 missing values (-1)
_____________________________________________________
* 152 subjects missing serum cholesterol at BOTH exams

*/

* USER: Point Stata to correct path for fram_data.txt file
*
*      ... use Set Working Folder (from File menu) 


*  --------------  Stata ----------------

clear

infile lexam dead cause cexam dead_chd dead_cva dead_ca not_chd sex age ht wt scl1 scl2 dbp sbp mrw smok using "fram_data.txt"
recode age    -1=.
recode  ht    -1=.
recode  wt    -1=.
recode  dbp   -1=.
recode  age   -1=.
recode  scl1  -1=.
recode  scl2  -1=.
recode  mrw   -1=.
recode  smok  -1=.
label variable lexam     "last exam"
label variable cause     "cause of death"
label variable cexam     "exam chd dx-ed"
label variable dbp       "diastolic BP"
label variable sbp       "systolic BP"
label variable scl1      "serum cholesterol-exam1"
label variable scl2      "serum cholesterol-exam2"
label variable smok      "cigarettes/day"
label variable mrw       "Metropolitan Relative Weight"


* for survival analysis  [ put death 1 year before 'last' exam ]  

gen fu_year = 2*(lexam -1) - dead 

* for NEW heart disease
 
gen t_newchd = (cexam>1) * ( 2*(cexam -1) - 1)  + (cexam==0) * fu_year
gen i_newchd = cexam > 1
replace t_newchd =.  if cexam==1 
replace i_newchd =.  if cexam==1


* indicator variables for male and female 

gen i_male    = (sex == 1)
gen i_female  = (sex == 2)

drop sex

* serum cholesterol 

gen chol = (scl1 + scl2 ) / 2
replace chol = scl2 if scl1==.
replace chol = scl1 if scl2==.

* for survival analysis

stset fu_year, failure(dead)

stcox i_male

stcox, nohr